Published Paper Details:

This study investigates the development of advanced feature engineering tools and techniques to uncover potential associations between polycystic ovary syndrome (PCOS) and gynecological cancers, such as ovarian cancer. By leveraging variables derived from ultrasound imaging and metabolic data, this research establishes a systematic approach to feature extraction, transformation, and selection for predictive modelling. Key ultrasound variables, such as follicle count and ovarian volume, are integrated with metabolic indicators, including glucose levels and hormonal profiles, to construct a comprehensive dataset. These features are then processed using methodologies like recursive feature elimination, correlation analysis, and principal component analysis (PCA) to identify the most significant predictors. Machine learning models, including logistic regression and random forests, are trained on the engineered features to evaluate the predictive accuracy and robustness of the approach. The results highlight the pivotal role of combining multimodal datasets in achieving high predictive performance, with random forests achieving an F1-score of 0.87. Furthermore, this research emphasizes the importance of feature engineering in medical diagnostics, offering insights into the complex interrelations between PCOS and ovarian cancer. The findings advocate for continued advancements in data integration and model development to support personalized healthcare interventions.